JP2019061424A - Classification device for referendum ballot paper - Google Patents

Abstract

To provide a classification device for referendum ballot paper capable of classifying a plurality of referendum ballot papers 2 collectively even when a plurality of voting themes coexist.SOLUTION: A classification device for referendum ballot paper includes a first voting theme determination part 30, a second voting theme 40, a written content determination part 50, and a classification part 60. The first voting theme determination part 30 determines a voting theme on the basis of color information of a ballot paper 2. The second voting theme determination part 40 determines a voting theme on the basis of image information of an entry face of the ballot paper 2. The written content determination part 50 determines written contents on the basis of the image information of the entry face of the ballot paper 2. The classification part 60 classifies the referendum ballot paper 2 according to the contents of each voting theme, on the basis of the voting theme determined by the first voting theme determination part 30, the voting theme determined by the second voting theme determination part 40, and the written contents determined by the written content determination part 50.SELECTED DRAWING: Figure 3

Description

  The present invention relates to a sorting apparatus for a referendum ballot.

  There is known an apparatus for reading a candidate name and the like entered on an election ballot and classifying the ballot (for example, Patent Document 1). Such a ballot paper classification apparatus picks up an image information by imaging the writing surface of the ballot paper (the surface on which the voters fill in the names of candidates etc.) with a scanner etc., and generates image information from this image information. The contents of the entry are read by image recognition, and the ballot paper is classified based on the read result.

Unexamined-Japanese-Patent No. 2004-5022

  In recent years, the possibility of revising the Constitution of Japan has become a reality. According to the Act on the Amendment Procedure of the Constitution of Japan (Law 51 of May 18, 2007), the entry column of the ballot paper for the referendum (“National ballot paper”) has “Agreed” and “Disagree” in advance. The letter is printed, and if the voter agrees to the constitutional amendment, the letter "Agree" is enclosed by ○, and if the voter disagrees, the letter "Averse" is circled. There is.

  In addition, it is also conceivable to vote on multiple voting themes such as "with regard to abolition of ● □ □" and "about the specification of ● ● △" on the same schedule, in which case dedicated for each voting theme The ballots will be able to complete multiple referendums. As described above, in the case where referendums of a plurality of voting themes are performed on the same schedule, it is convenient if the referendum papers of different voting themes can be collectively applied to the classification device.

  This invention was made to solve such problems, and it is possible to classify a plurality of referendum papers collectively even if there are a plurality of voting themes, and the classification of the referendum ballot It aims at providing an apparatus.

  In order to solve the above-mentioned problems, a classification device of a referendum according to the present invention comprises: a first determination unit which determines a voting theme of the referendum on the basis of color information of the referendum; A second determination unit that determines a voting theme of the referendum sheet based on image information of the sheet, and a third determination unit that determines the described contents of the referendum sheet based on image information of the referendum sheet And the voting theme determined by the first determination unit, the voting theme determined by the second determination unit, and the description content determined by the third determination unit. And a classification unit that classifies the voting theme according to the description content.

  The image forming apparatus further comprises an image sensor that irradiates a plurality of light of different colors onto the referendum paper, receives a plurality of reflected lights, measures each light reception amount of the plurality of reflected lights, and generates image information. Alternatively, color information may be generated based on the amount of light received by the plurality of reflected lights, and the voting theme may be determined based on the color information.

  The second determination unit extracts a voting theme area on which a voting theme is printed from the image information, and determines a voting theme based on either or both of pixel information and character information generated from the voting theme area. The third determination unit may extract a description area to be entered by the voter from the image information, and determine the description content of the description area based on pixel information generated from the description area.

  According to the classification apparatus for the referendum sheet of the present invention, even when a plurality of voting themes are mixed, it is possible to classify a plurality of referendum sheets collectively.

It is a figure showing the composition of the appearance of the classification device of the referendum paper according to the embodiment of the present invention. It is a schematic diagram which shows the internal structure of the classification device which concerns on embodiment of this invention. It is a schematic diagram which shows the structure of the arithmetic processing unit in the classification apparatus which concerns on embodiment of this invention. It is a figure which illustrates the format of a national ballot. It is a schematic diagram which shows the structure of a 1st voting theme discrimination | determination part in the classification apparatus which concerns on embodiment of this invention. It is a schematic diagram which shows the structure of the 2nd voting theme discrimination | determination part in the classification apparatus which concerns on embodiment of this invention. It is a schematic diagram which shows the structure of the description content discrimination | determination part in the classification device which concerns on embodiment of this invention. It is a flow chart which shows processing of learning mode in a classification device concerning an embodiment of this invention. It is a figure which illustrates the reference | standard mask pattern produced | generated by the classification | category apparatus which concerns on embodiment of this invention. It is a figure which illustrates the mask pattern for writing areas which is generated by the classification device according to the embodiment of the present invention. It is a flow chart which shows processing of classification mode in a classification device concerning an embodiment of this invention. It is a flow chart which shows processing of classification mode in a classification device concerning an embodiment of this invention. It is a flow chart which shows processing of classification mode in a classification device concerning an embodiment of this invention. It is a figure which illustrates the discrimination | determination process of the description content of the national ballot paper in the classification apparatus which concerns on embodiment of this invention. It is a figure which shows an example of the entry content by the voter in a referendum paper. It is a figure which shows an example of the entry content by the voter in a referendum paper. It is a figure which shows an example of the entry content by the voter in a referendum paper. It is a figure which shows an example of the entry content by the voter in a referendum paper. It is a figure which shows an example of the entry content by the voter in a referendum paper. It is a figure which shows an example of the entry content by the voter in a referendum paper.

Hereinafter, an embodiment of the present invention will be described based on the attached drawings.
Embodiment.
The ballot classification device classifies the ballot based on the voting theme of the ballot and the contents written by the voters. As shown in FIG. 1, the classification device is composed of a main body 1 and an extended stacker unit 8. The main body 1 of the classification device is provided with a hopper 3 for storing a plurality of national voting papers 2 to be classified, and a transport path 4 composed of a plurality of rollers 4 a for transporting the national voting papers 2.

  Further, the main body 1 of the classification device irradiates a plurality of lights of different colors onto the national voting paper 2 to receive a plurality of reflected lights, measures each light reception amount of the plurality of reflected lights, and generates image information. A sensor 5, five stackers 6 a to 6 e for classifying and storing the national voting paper 2 based on the voting theme and the described contents, and a monitor 7 for displaying the classification result.

  The extension stacker unit 8 is an extension unit which can be separated from the main body 1, and includes an extension conveyance path 9 constituted by a plurality of rollers 9a and eight extension stackers 10a to 10h.

  In the stackers 6a to 6e provided in the main body 1 of the sorting apparatus and the extended stackers 10a to 10h provided in the extended stacker unit 8, national voting papers 2 classified as described later are accommodated. As an example, in this embodiment, in the stackers 6a to 6d of the main body 1, the affirmative vote, the negative vote, the white vote, the invalid vote and the question vote in the first voting theme (“About abolition of ● □□”) The ballot paper accommodated therein and rejected as being unclassifiable is accommodated in the stacker 6e. In addition, the expansion stackers 10a to 10d of the expansion stacker unit 8 accommodate a positive vote, a negative vote, a white vote, an invalid vote, and a questionnaire on the second voting theme (for the specification of “● ΔΔ”).

  Next, as shown in FIG. 2, the image sensor 5 includes a main body 11, and three LEDs 12 a to 12 c are provided inside the main body 11. The LEDs 12 a to 12 c emit light of different colors, specifically, the LED 12 a emits red light, the LED 12 b emits green light, and the LED 12 c emits blue light.

  The main body portion 11 has a housing portion 11a which is a space opened toward the national ballot paper 2 conveyed along the conveyance path 4, and the LEDs 12a to 12c are provided in the housing portion 11a. It is supported by the light guide plate 13. The light guide plate 13 is a member formed of a transparent material, and each light emitted from the LEDs 12a to 12c is irradiated to the national voting paper 2 through the light guide plate 12 as shown by the arrow B1.

  Further, the main body 11 has a housing 11b which is a space opened toward the national ballot paper 2. A light guide plate 14 made of a transparent material is provided also in the housing 11b. A light receiving sensor 15 is provided at an end of the light guide plate 14 located on the back side of the housing portion 11 b. The red, green and blue lights emitted by the LEDs 12a to 12c are reflected by the national ballot paper 2 as indicated by the arrow B2, and the reflected lights reach the light receiving sensor 15 through the light guide plate 14.

  The light receiving sensor 15 receives each reflected light from the national voting paper 2 to measure each light receiving amount and generates image information, and the measured light receiving amount and the generated image information are described in the arithmetic processing unit 20 described next. Output to Each of the LEDs 12 a to 12 c is electrically connected to the control unit 16, and the control unit 16 controls start and stop of light irradiation.

(Operation processing unit 20)
As schematically shown in FIG. 3, an arithmetic processing unit 20 constituted by a microcomputer or the like is provided in the main body 1 of the classification device. The arithmetic processing unit 20 includes a first voting theme determination unit 30, a second voting theme determination unit 40, a written content determination unit 50, and a classification unit 60.

  The first voting theme determination unit 30 determines the voting theme of the national voting paper 2 from each light reception amount input from the image sensor 5. In detail, as illustrated in FIG. 4, the referendum ballot 2 is prepared with a dedicated ballot for each voting theme such as “with regard to abolition of ● □□” and “with regard to the specification of ●● ΔΔ” The color of the paper is set for each voting theme. In the example of FIG. 4, the ballot paper of “abolition of ● □□” in (A) is pink, and the ballot paper of “indication of ●● ΔΔ” in (B) is yellowish . The first voting theme determination unit 30 generates color information of the national voting paper 2 based on each light receiving amount input from the image sensor 5, and determines a voting theme of the national voting paper 2 based on the color information.

  The second voting theme determination unit 40 extracts, from the image information generated by the image sensor 5, an area (voting theme area 2a) on which the voting theme in the national voting paper 2 is printed, and generates the voting theme area 2a. Based on the pixel information and the character information, the voting theme of the national voting paper 2 is determined.

  The written content determination unit 50 extracts areas (written areas 2 b and 2 c) to be filled in by the voter in the national ballot paper 2 from the image information input from the image sensor 5 and generates the areas from the written areas 2 b and 2 c. Based on the pixel information, the described contents of the referendum ballot 2 are determined.

  The classification unit 60 is based on the voting theme determined by the first voting theme determination unit 30, the voting theme determined by the second voting theme determination unit 40, and the description content determined by the description content determination unit 50. Classify the referendum ballot 2 according to the contents described for each voting theme.

  Next, detailed configurations of the first voting theme determination unit 30, the second voting theme determination unit 40, and the description content determination unit 50 will be described in order.

(First voting theme determination unit 30)
As schematically shown in FIG. 5, the first voting theme determination unit 30 includes a color information generation unit 31, a storage unit 32, and a determination unit 33.

  The color information generation unit 31 receives each light reception amount measured by the image sensor 5, that is, each light reception amount of each reflected light from the national voting paper 2 for each of the red, green, and blue lights irradiated by the LEDs 12a to 12c. It is input. The color information generation unit 31 generates color information of the national voting paper 2 based on the red, green and blue light reception amounts.

  The storage unit 32 stores each color information of the referendum ballot 2 for learning corresponding to each voting theme as reference color information Cr1, Cr2,. . . Remember as.

  The determination unit 33 determines the color information (evaluation color information Ce) of the national ballot paper 2 of the classification target generated by the generation unit 31 and the reference color information Cr1, Cr2,. . . And determine the voting theme of the referendum ballot 2 to be classified.

(2nd voting theme determination unit 40)
As schematically shown in FIG. 6, the second voting theme determination unit 40 includes an area extraction unit 41, a reference information generation unit 42, a storage unit 43, an evaluation information generation unit 44, and a determination unit 45. It is configured.

  The area extracting unit 41 is an area on which voting themes such as “About abolition of ● □□” and “About specifying Δ ●” are printed from the image information of the national voting paper 2 generated by the image sensor 5 ( Extract the voting theme area 2a).

  Based on the pixel information of each voting theme area 2a of the referendum sheet 2 for learning corresponding to each voting theme, the reference information generating unit 42 generates reference mask patterns Mr1, Mr2,. . . Generate In addition, the reference information generation unit 42 performs character recognition processing on each voting theme area 2a of the referendum sheet 2 for learning corresponding to each voting theme, whereby reference character strings Sr1, Sr2,. . . Generate

  The storage unit 43 stores reference mask patterns Mr1, Mr2,. . . And reference strings Sr1, Sr2. . . Remember.

  The evaluation information generation unit 44 generates evaluation pixel information Pe based on the pixel information of the voting theme area 2a of the national ballot paper 2 to be classified. Further, the evaluation information generation unit 44 generates an evaluation character string Se by performing character recognition processing on the voting theme area 2a of the national ballot paper 2 to be classified.

  The determination unit 45 compares the evaluation pixel information Pe of the classification target national ballot paper 2 generated by the evaluation information generation unit 43 with the reference mask patterns Mr1 and Mr2.m stored in the storage unit 43. . . The evaluation character string Se of the classification target national ballot paper 2 generated by the evaluation information generation unit 44, and the reference character strings Sr1, Sr2,. . . On the basis of the comparison result with the above, the voting theme of the national ballot paper 2 to be classified is determined.

(Description determination unit 50)
As schematically shown in FIG. 7, the written content determination unit 50 includes an area extraction unit 51, a mask generation unit 52, a storage unit 53, and a determination unit 54.

  From the image information of the national ballot paper 2 generated by the image sensor 5, the area extraction unit 51 is an area where the character strings “Agreement” and “Disagreement” are printed in advance and an entry is made by the voter , 2c).

  The mask generation unit 52 generates mask patterns Mb and Mc based on pixel information of the writing areas 2b and 2c of the unwritten national ballot paper 2 for learning.

  The storage unit 53 stores the mask patterns Mb and Mc generated by the mask generation unit 52.

  The determination unit 54 applies the mask patterns Mb and Mc stored in the storage unit 53 to the pixel information of the described areas 2 b and 2 c in the national voting paper 2 to be classified and extracted by the area extraction unit 51. To determine the content written by the voter.

  Next, the process performed by the classification device for the referendum according to this embodiment will be described.

  The referendum sorting apparatus has two types of processing modes, "learning mode" and "classification mode". In the learning mode, one reading national ballot paper 2 for learning corresponding to each voting theme is read one by one, and those features are stored as reference information. In the classification mode, evaluation information is generated from each of a plurality of referendums 2 filled by the voter, and each referendum 2 is voted based on the evaluation information and the reference information stored in the learning mode. Classify based on the theme and description content. The details of the processing in the learning mode and the classification mode will be sequentially described below.

(Learning mode)
In order to start the processing of the learning mode, the user takes, as a referendum for learning, one uncompleted reference ballot 2 corresponding to each voting theme and places it on the hopper 3 and operates the switch not shown. Instructs the start of the learning mode.

  When instructed to start the learning mode, the classification device conveys each of the national voting papers 2 for learning to the image sensor 5 by driving the rollers 4a constituting the conveyance path 4 to the flowchart of FIG. Do the processing shown.

  First, the control unit 16 (see FIG. 2) of the image sensor 5 sequentially starts and stops the irradiation of light by each of the LEDs 12a to 12c for each of the national voting papers 2 for learning, and the light receiving sensor 15 of the image sensor 5 The reflected light from the national voting paper 2 to the light of each color sequentially emitted by each of the LEDs 12a to 12c is sequentially received to measure each received light amount (S101). The image sensor 5 outputs the measured amounts of light reception to the first voting theme determination unit 30 of the arithmetic processing unit 20.

  The generation unit 31 of the first voting theme determination unit 30 generates reference color information Cr1, Cr2,. . . To generate these reference color information Cr1, Cr2,. . . Are stored in the storage unit 32 (S102). Here, Cr1 is reference color information of the first learning national ballot 2 for learning, Cr2 is reference color information of the second learning national ballot 2 for learning, and so forth.

  Further, each reference color information Cr is generated as relative intensities of the light reception amount of red reflected light, the light reception amount of green reflected light, and the light reception amount of blue reflected light. In the example of FIG. 4, the national ballot paper 2 of (A) is pink, and the relative intensity of the amount of received light of the reflected light of red is the highest. In addition, the national ballot paper 2 of (B) is a yellowish color, and the relative intensity of the amount of received green reflected light is the highest.

  Next, the image sensor 5 generates the binarized image information of the entry surface of each of the national voting papers 2 for learning (S103). The image sensor 5 outputs the generated image information to the second voting theme determination unit 40 and the written content determination unit 50 of the arithmetic processing unit 20.

  The area extraction unit 41 of the second voting theme determination unit 40 extracts the voting theme area 2a from the image information of each of the national voting papers 2 for learning (S104). In the example of FIGS. 4 (A) and 4 (B), a rectangular area surrounded by a broken line including a character string “For elimination of ● □□” and “About specification of ●● ΔΔ” is extracted as the voting theme area 2a. Be done.

  Further, the coordinates of the voting theme area 2a are stored for use in the later determination mode. As a method of extracting the voting theme area 2a, an area including a character string representing a voting theme may be specified by performing character recognition processing on the entire image information. May be displayed on the screen to allow the user to specify an area. Moreover, in this embodiment, the coordinates of the voting theme area 2a are assumed to be the same in all kinds of national voting papers 2.

  Next, the reference information generation unit 42 performs expansion processing and inversion processing on the pixel information of the voting theme area 2a extracted in step S104, so that the reference mask patterns Mr1, Mr2,. . . Are generated and these reference mask patterns are stored in the storage unit 43 (S105). In the examples of FIGS. 4A and 4B, reference mask patterns as shown in FIGS. 9A and 9B are generated.

  Further, the reference information generation unit 42 performs character recognition processing on the pixel information of the voting theme area 2a extracted in step S104, thereby extracting character strings included in the voting theme area 2a, and referring to these The character string is stored in the storage unit 43 (S106). In the example of FIG. 4 (A) (B), the character strings “for the abolition of ● □□” and “for the specification of ●● ΔΔ” are respectively extracted, and these are stored in the storage unit 43 as a reference character string. Ru.

  Next, the area extraction unit 51 of the written content determination unit 50 extracts the written areas 2b and 2c from the image information of the first learning national ballot paper 2 (S107). In the example of FIG. 4A, rectangular regions surrounded by broken lines including character strings “agree” and “opposite” are extracted as the description regions 2b and 2c, respectively. In this embodiment, it is assumed that the character strings ("Yes" and "No") printed in advance in the entry areas 2b and 2c are the same in all types of the referendum papers 2.

  In addition, the coordinates of the described areas 2b and 2c are stored for use in the classification mode described later. In addition, as a method of extracting the description areas 2b and 2c, by performing character recognition processing on the entire image information, it is possible to respectively specify areas including character strings of "yes" and "opposite". Image information may be displayed on the monitor 7 to allow the user to designate an area. Moreover, in this embodiment, the coordinates of the description areas 2b and 2c are assumed to be the same in all types of the referendum papers 2.

  Next, the generation unit 52 performs expansion processing and inversion processing on the pixel information of the writing regions 2 b and 2 c extracted in step S 107 to generate mask patterns Mb and Mc for writing regions, The mask pattern is stored in the storage unit 53 (S108). In the example of FIG. 4A, a mask pattern as shown in FIG. 10 is generated.

  By the above processing performed in the learning mode, the storage unit 32 of the first voting theme determination unit 30 stores reference color information Cr1, Cr2,. . . Are stored, and the storage unit 43 of the second voting theme determination unit 40 stores reference max patterns Mr1, Mr2,. . . And reference strings Sr1, Sr2,. . . Are stored, and in the storage unit 53 of the written content determination unit 50, mask patterns Mb and Mc for the writing area are stored.

(Classification mode)
When the processing of the learning mode is completed, the user arranges the plurality of referendums 2 filled by the voter in the hopper 3 and instructs the start of the classification mode by an operation switch (not shown).

  When instructed to start the classification mode, the classification device conveys the national voting papers 2 to be classified to the image sensor 5 one sheet at a time by driving the rollers 4a constituting the conveyance path 4, and the respective national votes The process shown in the flowchart of FIG. 11 is performed on the sheet 2.

  First, the control unit 16 (see FIG. 2) of the image sensor 5 sequentially starts and stops the irradiation of light by the LEDs 12a to 12c for the national voting paper 2 to be classified, and the light receiving sensor 15 of the image sensor 5 Reflected light from the national ballot paper 2 to the light of each color sequentially emitted by each of the LEDs 12a to 12c is sequentially received to measure each received light amount (S201). The image sensor 5 outputs the measured amounts of light reception to the first voting theme determination unit 30 of the arithmetic processing unit 20.

  The color information generation unit 31 of the first voting theme determination unit 30 generates evaluation color information Ce based on each light receiving amount of the national ballot paper 2 to be classified (S202). The determination unit 33 determines the evaluation color information Ce generated by the color information generation unit 31, the reference color information Cr1, Cr2,. . . (S203), and it is checked whether there is reference color information that matches or approximates the evaluation color information Ce (S204).

  In step S204, when there is reference color information that matches or approximates the evaluation color information Ce (S204 = YES), the voting theme of the national voting paper 2 to be classified corresponds to the reference or matching reference color information. It is determined that it is a voting theme to be played (S205). On the other hand, if there is no reference color information that matches or approximates the evaluation color information Ce (S 204 = NO), the referendum ballot 2 to be classified is rejected as an unclassifiable mixed vote (S 206) . As described above, the rejected ballot paper is stored in the stacker 6 e of the main body 1.

  Next, the image sensor 5 generates binarized image information of the entry surface of the national voting paper 2 to be classified (S207). The image sensor 5 outputs the generated image information to the second voting theme determination unit 40 and the written content determination unit 50 of the arithmetic processing unit 20.

  The area extraction unit 41 of the second voting theme determination unit 40 extracts the voting theme area 2a from the image information based on the coordinates stored in the learning mode (S208).

  The evaluation information generation unit 44 extracts each pixel of the voting theme area 2a to generate evaluation pixel information Pe (S209a). Further, the evaluation information generation unit 44 performs character recognition processing on the voting theme area 2a, thereby extracting a character string included in the voting theme area 2a and setting it as an evaluation character string Se (S209b).

  The determination unit 45 compares the evaluation pixel information Pe generated at step S209a with the reference mask patterns Mr1, Mr2,. . . Are sequentially applied, and the number of remaining pixels is checked (S210a). That is, the logical product (AND) of the corresponding pixels of the evaluation pixel information Pe and the reference mask pattern is calculated, and the number of pixels having a value of 1 is checked.

  In step S210a, when the voting theme of the national voting paper 2 to be classified matches the voting theme corresponding to the reference mask pattern, as a result of applying the reference mask pattern to the evaluation pixel information Pe, the value is 1 Of the pixels hardly remain.

  On the other hand, in step S210a, when the voting theme of the national voting paper 2 to be classified is different from the voting theme corresponding to the reference mask pattern, the result of applying the reference mask pattern to the evaluation pixel information Pe A large number of pixels with a value of 1 remain.

  The determination unit 45 further evaluates the evaluation character string Se extracted in step S 209 b and the reference character strings Sr 1, Sr 2,. . . Are sequentially compared, and it is checked whether there is a reference character string that matches the evaluation character string Se (S210b).

  The determination unit 45 determines that there is a reference mask pattern in which the number of remaining pixels as a result of ANDing the evaluation pixel information Pe is equal to or less than a predetermined value, and a reference character string that matches the evaluation character string Se. It is checked whether or not it is (S211).

  In step S211, when there is a reference mask pattern and a reference character string that satisfy such conditions (S211 = YES), the voting theme of the national voting paper 2 to be classified is the reference mask pattern and the reference that satisfies the condition. It is determined that it is a voting theme corresponding to the character string (S212).

  On the other hand, when such reference pixel information and reference character string do not exist (S211 = NO), the national ballot paper 2 to be classified is regarded as an unclassifiable mixed vote and rejected (S213).

  In the processing of steps S209 to S213, the discrimination accuracy is improved by discriminating the voting theme based on the pixel information in steps S209a to 210a and the discrimination of the voting theme based on the character information in steps S209b to S210b. Although this is intended, in order to shorten the processing speed and save computational resources, either discrimination based on pixel information or discrimination based on text information may be performed.

  Next, the area extraction unit 51 of the written content determination unit 50 extracts the written areas 2b and 2c from the image information based on the coordinates stored in the learning mode (S214).

  The determination unit 54 extracts each pixel of the description areas 2b and 2c to generate pixel information, and applies the mask patterns Mb and Mc stored in the storage unit 53 to the pixel information (S215).

  For example, in the case where the description regions 2b and 2c are as shown in FIG. 12A, applying the mask patterns Mb and Mc in step S215 provides pixel information as shown in FIG. 12B. . That is, by applying the mask patterns Mb and Mc to the entry areas 2b and 2c in which entry is made by the voter, the character string printed in advance in the entry areas 2b and 2c is removed, and the entry is made by the voter Only the part that has been However, as a result of applying the mask patterns Mb and Mc, a part of the circle drawn by the voter is missing.

  Next, the determination unit 54 performs expansion processing on the pixel information after applying the mask patterns Mb and Mc in step S215 to generate pixel information as shown in FIG. 12C (S216). . In FIG. 12 (C), the circle drawn by the voter is restored.

  Subsequently, the determination unit 54 checks whether or not a predetermined number or more of pixels having a pixel value of 1 exist in the pixel information generated in step S216 (S217). Here, when the number of pixels having a pixel value of 1 does not exist a predetermined number or more (S217 = NO), it is determined that the white sheet is as shown in FIG. 13 (S218).

  On the other hand, when the number of pixels having a pixel value of 1 is greater than or equal to a predetermined number (S217 = YES), the determination unit 54 corresponds to “Yes” and “No” in the pixel information generated in step S216. In the areas 2b and 2c, it is checked whether a predetermined number or more of pixels having a pixel value of 1 exist (S219). Here, when the number of pixels having a pixel value of 1 is greater than or equal to a predetermined number in both of the regions 2b and 2c (S219 = YES), as shown in FIGS. It determines that it is an invalid vote in which the entry is made in both of the fields (S220).

  On the other hand, when the number of pixels having a pixel number of 1 does not exist in a predetermined number or more in both of the regions 2b and 2c (S219 = NO), the determination unit 54 determines whether the left or right corresponds to It is checked whether there is a closed circle in the area (S221). Here, if there is no closed circle in any of the regions (S221 = NO), as shown in FIGS. It is determined that the question is not filled in (S222).

  On the other hand, when a closed circle is entered in the area of either "Agree" or "Disagree" (S221 = YES), it is determined that the vote is affirmative or negative (S223). That is, if a closed circle is filled in the area 2b corresponding to "yes", it is judged as an affirmative vote, and if a closed circle is filled in the area 2c corresponding to "opposite" It is determined that the vote is negative.

  Finally, the classification unit 60 determines the voting theme determined by the first theme determination unit 30 in step S205, the voting theme determined by the second theme determination unit 40 in step S212, and the described content determination in steps S217 to S223. Based on the description content determined by the section 50, the classified voter's ballot paper 2 is classified as a white vote, an invalid vote, a question vote, a positive vote or an negative vote for each voting theme (S224).

  If the voting theme determined by the first theme determination unit 30 is different from the voting theme determined by the second theme determination unit 40 in step S224, the national voting paper 2 of the classification target is classified Reject it as an impossible mixed vote.

  By the above processing performed in the classification mode, it is possible to classify the national ballot paper 2 to be classified into any of white votes, invalid votes, question votes, and positive or negative votes for each voting theme. And, as mentioned above, in this embodiment, the affirmative vote, the negative vote, the white vote, the invalid vote and the question vote in the first voting theme (“About the abolition of ● □□”) 6d, and in the second voting theme (for the description of “● 明 記”), the affirmative vote, the negative vote, the white vote, the invalid votes and the questionnaires are respectively received in the extended stackers 10a to 10d of the extended stacker unit 8 Be done.

  As described above, the classification device according to the embodiment of the present invention includes the first voting theme determination unit 30, the second voting theme 40, the written content determination unit 50, and the classification unit 60.

  The first voting theme determination unit 30 determines a voting theme of the referendum sheet 2 based on color information of the referendum sheet 2. The second voting theme determination unit 40 determines the voting theme of the referendum sheet 2 based on the image information of the entry surface of the referendum sheet 2. The written content determination unit 50 determines the written content of the referendum sheet 2 based on the image information of the entry surface of the referendum sheet 2.

  The classification unit 60 is based on the voting theme determined by the first voting theme determination unit 30, the voting theme determined by the second voting theme determination unit 40, and the description content determined by the description content determination unit 50. The referendum ballot 2 is classified according to the content of each voting theme.

  According to the above-mentioned feature, in the classification apparatus of the referendum paper according to the embodiment of the present invention, even when plural voting themes are mixed, plural referendum papers 2 can be classified at one time. In the above embodiment, an example is shown in which one expansion stacker unit 8 is connected to the main body 1, but a plurality of expansion stacker units 8 can also be connected to the main body 1. Therefore, it is possible to cope with a large number of voting themes by increasing the number of the extended stacker units 8 to be connected.

  2 national ballot paper, 5 image sensor, 30 first voting theme determination unit (first determination unit) 40 second voting theme determination unit (second determination unit) 50 description content determination unit (third determination unit ), 60 classification department.

Claims (3)

Classification machine for referendum ballot,
A first determination unit that determines a voting theme of the referendum based on color information of the referendum;
A second determination unit that determines a voting theme of the referendum sheet based on image information of the referendum sheet;
A third determination unit that determines the described content of the referendum sheet based on the image information of the referendum sheet;
The voting theme determined by the first determination unit, the voting theme determined by the second determination unit, and the description described based on the third determination unit. A classification device for a referendum paper, comprising: a classification unit that classifies the referendum paper according to the contents for each voting theme. The national voting paper is further provided with an image sensor that emits a plurality of light of different colors to receive a plurality of reflected lights, measures each light reception amount of the plurality of reflected lights, and generates image information.
The classification of a national voting paper according to claim 1, wherein the first discrimination unit generates color information based on each light reception amount of the plurality of reflected lights, and discriminates the voting theme based on the color information. apparatus. The second determination unit extracts a voting theme area on which a voting theme is printed from the image information, and the voting theme is extracted based on either or both of pixel information and character information generated from the voting theme area. To determine
The third determination unit extracts a description area to be entered by a voter from the image information, and determines the description content of the description area based on pixel information generated from the description area. National ballot classification device described in.

Images